Network Card for Gamers - Uses Linux to Reduce Lag

Cujo writes "The folks at GDHardare have an interview with Bigfoot Networks discussing the pending release of their Killer Network Card which is said to greatly reduce in-game latency. According to the Interview, this card uses a Linux-based subsystem to do its magic."

Comments

Yes.

[Spazntwich]

I'm sure another layer of abstraction to the network is exactly what gamers need to reduce lag.

Overloaded and slow routers will say, "Whoah, his network card RUNS LINUX. I'll shuffle these packets through more quickly."

I'd believe their hype more if we already had an openly tiered internet and these guys gave you a free year's subcription to the top tier with purchase of the card.

Re:Yes.

[CyberBill]

A lot of games are TCP based... World of Warcraft for example. Even games that arent exactly TCP are typically a reliable messaging system on top of UDP that pretty much mimics TCP.

With that said, I cant see how this network card could reduce your latency by more than 1ms or 2ms round trip. Latency isnt introduced because your PC is stupid, its introduced because you're waiting the time it takes for packets to travel to your ISP, to its ISP, to its ISP, down to its child, down to its child, and back to some other PC, and having to interact with the 20 routers, gateways, and switches along the way. Most switches use something called Hold and Forward (I might have the name wrong...) which listens for the whole packet, reads the header information, and then passes it along, rather than writing the bits as they come in like a hub does... (Please dont read into this and think hubs are better :P )

Re:Yes.

[russ1337]

....and probably bypasses any software firewall on your machine at the same time... how long till there is an exploit to get the network card to trigger 'gaming mode' for a worm... my bet is 2 days from release...

Re:Yes.

[Jherek Carnelian]

Even games that arent exactly TCP are typically a reliable messaging system on top of UDP that pretty much mimics TCP.

With that said, I cant see how this network card could reduce your latency by more than 1ms or 2ms round trip.

Given the constraints - TCP and various homegrown reliable protocols on top of UDP, it isn't too hard to come up with some options to improve latency. But they all involve violating the RFCs.

First you have to wrap head around one important factor that can absolutely kill latency for any transport with guaranteed delivery -- packet loss. Packet loss means you have to discover what packets were lost and then retransmit them - those two steps can easily introduce delays on the order of seconds.

So one trick would be to pre-send the retransmits. Send duplicate packets spaced apart by a few miliseconds. If the other end receives multiple copies of the same packet, it will silently discard any extras - but if one copy gets lost en route, the other packet might still make it through, thus eliminating the whole timeout/retransmit cycle. It should be possible to do this for both TCP and UDP.

However, doing something like that is very unfriendly because it wastes resources. The primary reason packets get lost en route is because of bandwidth saturation. So, if you double or triple your traffic you are just making the problem worse. If you are the only one out of thousands who "breaks the rules" you will probably get away with it and probably even benefit from it since packet loss will be a somewhat even distribution among all traffic, so chances are if one of your packets gets dropped the copy won't get dropped - instead someone else's packet gets dropped.

But if a significant minority of users were to do the same thing, it would probably result in a complete collapse of any usuable bandwidth. Which is exactly the kind of thing I would expect a bunch of MBA's to come up with.

Re:Yes.

[mrbcs]

IMNSHO, most latency is caused by "CRAPPY $5 onboard NETWORK CARDS!"

Get a 3com NIC and be done with it. I've seen a machine getting pings of 200 to 300 and when I switched the crap card to a real 3com the ping times went down to 50.There is a reason that 3com charges $35 for their cards... they work well. I use them exclusively in my network.

Shitty hardware always causes problems.

Pricey

[HeWhoRoams]

Pre order cost is $280. You'll see a better FPS increase spending that on a graphics card, RAM, or some groceries for 6 months.

Oooo... Killer

[snowgirl]

OMG, they named it the "KillerNIC"? Like, does this kind of advertising actually work?

"This NIC is so hardcore it KILLED SOMEONE!"

I can just imagine their second version coming with a muzzle a la Silence of the Lambs.

Is this really needed?

[adamwright]

As a small test, I ran up Quake 3 on it's highest settings, and had it play back a reasonably heavy demo. Now, Quake3 isn't the most modern of games, but it can still peg a CPU at 100%. Then, I found the latency to my router.

Pinging 192.168.0.1 with 32 bytes of data:

Reply from 192.168.0.1: bytes=32 time1ms TTL=255
Reply from 192.168.0.1: bytes=32 time1ms TTL=255
Reply from 192.168.0.1: bytes=32 time1ms TTL=255

Assuming this product entirely eliminates all latency on the first hop (impossible), that's a net gain of 1ms.

The entire concept of these FNApps also strikes me as a route to evil; I heard a subtext of "Now, even the most clueless Windows gamer with too much money can run packet scanning cheating tools with no chance of detection!".

I'm placing this one firmly in the "Snake oil" bin, based on this interview.

Retarded,

[u16084]

If your ISP sucks ass, a $250 lan card is not going to help.

So it's a QoS Network Card?

[tlhIngan]

The only thing I can guess it needs Linux for is to do the routing and QoS services (see lartc.org)...

Then again, considering I get sub-1ms latencies across my network (only 100Mbps...), and this is with some rather pathetic equipment (Celeron system running Win2k), I fail to see how I can improve my 80ms ping with a better network card.

It seems that hardcore gamers are starting to become the computing equivalent of the "audiophile". From CRT displays that do 120hz refresh (do they notice the difference between 100 and 120, I wonder?) since LCDs that do 6ms are "too slow". Gaming mice that do 10k-dpi for ultra-precise positioning, videocards that cost the better part of a grand. And now, network cards that cut down microseconds or give you that extra frame per second. There's also keyboards, the gaming mousepad (though, some are nice for general use), and god knows what other accessories, doodads and other monster-cable-type things.

Re:So it's a QoS Network Card?

[slackmaster2000]

Yeah audiophiles are very easily fooled. I laugh every time I hear about digital technology being described in analog terms, which is terribly common in the audiophile realm. It's sort of understandable though because "digital" is so variable whenever we interface with the analog world. Not all AD/DA converters are created equal. Audio CD playback is very similar to analog technology in that it's designed play right through and mask errors which can result in high frequency distortion and various other artifacts that one might not expect from "digital" and how that term is marketed. It can be difficult to understand, so why wouldn't the quality of a USB cable be as important as the quality of a microphone cable or the weight of a speaker cable? Well, most of us would say "duh", but the second big problems is that these people can "hear" the difference! Ah the power of suggestion. You won't find a lot of double blind studies published by audiophile gear manufacturers.

Aside from $30,000 speaker cables (I shit you not), my most favorite audiophile product is a wooden knob that costs $500. "What does the wooden knob do, though?" Well, nothing on its own. It is in fact, a wooden knob. A knob made from wood. It sounds great when attached to your volume pot, though. Five hundred dollars please.

So today we learn about a network card that somehow reduces lag by implementing hardcore quality of service on an endpoint that is for all intents and purposes dedicated to a single application. Are gamers analogous to audiophiles? "Quake III is a lot more responsive now. My ping is about the same, but I can feel the difference." It must work though. I mean, who would make something and charge so much for it if it didn't actually work? :)

Re:Huh?

[forkazoo]

It's always been my understanding that the bigger bottlenecks are upstream of your NIC. I mean, my home network set up goes gigabit from my desktop to my hardware router, gigabit from my router to my gateway firewall, then gigabit (minus a few MTU) to my DSL modem, and after that the speed gets massively reduced and there's nothing I can do about it. My lan latency is practically non-existant.

Now, maybe I'm completely misunderstanding teh point of this NIC, but...

You are correct. The NIC isn't an appreciable source of latency. Right now, I ping'd a server on another subnet, and I averaged 0.3 ms latency. This is bog standard 100 Mb. Nothing the least bit fancy. That server might have a nice NIC of some sort, but this desktop certainly doesn't. And, that's hopping between subnets. Crossing between buildings over a T-1, with a few routers involved in about 5 ms. Pinging my home machine over the internet is abou 150 ms. So, assuming that of the .3 ms latency I have inside this building, none of it is due to the switch, and none of it is due to actual wire delay, then about half of the latency is my system, and half is from the server. So, my NIC is responsible for abou 0.15 ms of latency.

Now, assuming that I was playing a game with my home computer, moving to a NIC that cut the latency of my PC down by 2/3 (from .15 ms to .05 ms), I'd be shaving my total latency for the connection to 149.9 ms (from 150ms).

Which would improve my lag by .06%

No, dammit. You won't see a noticeable improvement from a lower latency NIC. There are probably a few microbenchmarks where you will get a phenomenal speedup. Gaming isn't one of those cases.

Re:Is it credible?

[snowgirl]

It's obvious that they're all about the business here.

"Powered by Lag and Latency Reduction (LLR) Technology"
"Future-Proof: Field Upgradeable"
"UltimatePing(tm)"
"MaxFPS(tm)"
"FNA(tm)"
"GameFirst(tm)"
"PingThrottle(tm)"

Seriously, who else but a marketting department would think that it's a good idea to trademark a name describing everything "new" that your product does? And the page is so full of TLAs (three letter acronyms) that you need a glossary to read it.

So, yes, I'd have to weigh in with everyone else, it's snakeoil. Basically, any product designed entirely by a marketting group is going to be snakeoil, and this definitely was.

I think you are making some invalid comparisions

[Sycraft-fu]

For example take the CRT thing. I own such a CRT, and it's not marketed to gamers, it's marketed to professionals. Why the refersh rate then? Simple function of it's ability to go super high resolution. The monitor is rated to do 2048x1536 @85Hz. To do that, takes some fast electron guns. Well, that ability implies higher refresh rates at lower resolutions. It can do over 200Hz at 800x600 because the resolution is so low. The point is to get extremely high resolutions at usable refresh rates. Also, in general, you want your device spec'd above what it's supposed to actually do. You don't want to run it at it's limits all the time.

Likewise the mouse thing is a little misinformed. Higher DPI cameras isn't worthless on an optical mouse. It lets it track on more uniform surfaces. No matter how uniform something looks, at some point it's uneven. Well, optical mice need uneveness to track, that's why they don't work on a mirror, or a really smooth surface, they can't track details. One way to make them track better is to up the DPI. The smaller details they see, the more uniform a surface can be. That's also the point behind using a laser. Since it is truly monochromatic light, just one frequency, it shows small details in a starker contrast that is lost with normal LED light.

Though there's certianly BS targeted at the gamer market, this being some of the BS, there's plenty of products with real legit reasons to be bought. Not everyone wants an experience that is "acceptable" or "works jsut good enough to get the job done." Doesn't mean they are wasting money on the things they buy. Yes a $50 used mountain bike will get me to work and back, but that doesn't mean that I'm wasting money on a deceant $600 street bike. It honestly does work better.

Ted Stevens and I believe the card works

[User 956]

So, yes, I'd have to weigh in with everyone else, it's snakeoil. Basically, any product designed entirely by a marketting group is going to be snakeoil, and this definitely was.

Everyone knows the internets is a series of tubes. Well, this card hurls your data through the tubes with such force that it can't possibly get stuck.

Make sure not to point the jack at anyone. You'll shoot someone's eye out.

Tytus (Bigfoot Networks) Responds

[Tytus_Bigfoot]

Hey everyone, I am the inventor of the Killer NIC. I will not try to hijack your thread, so this will be my only post. Thanks for the interest in our Killer Network Card. It has been my personal vision for years. A lot of very good questions have been raised here, and I think a lot of them are answered in our FAQ here: http://www.bigfootnetworks.com/FrequentlyAskedQues tions.aspx . If there are still questions, I would love to try to answer them at our sponsored community site: http://www.endlagnow.org/ELNForums/ Thanks, Tytus

Re:Is it credible?

[SoloFlyer2]

While the network card Might improve the network connectivity for that one person, it more than likely does it by hurting the performance of the the network as a whole...

IE see the ALOHA protocol...
http://en.wikipedia.org/wiki/Aloha_protocol#The_AL OHA_protocol

Changing the way that the network card responded so that rather than using a random backoff time it just retried staight away would decrease the network latency but if there were several of the cards on the network all trying to do the same thing the network would grind to a halt...

There are alot of protocols with backoffs etc which could be changed to improve the individuals connection but making them faster would be a BAD THING for the network as a whole!

but, there are also other things which could be done to improve network connectivity without hurting the network as a whole... just setting up some kind of QoS on the network card so that small packets like ping response would always get pushed to the front of the queue would improve the ping time signifigantly under non-idle situations...